Object Recognition Based on Modified Invariant Moments

We present a novel method for object recognition in noise free and noisy environments, based on modified invariant moments and minimum norm. First, the modified invariant moments of different objects are extracted by using invariant moments. Then the norms of feature vectors are computed by using norm theory of functional analysis. Finally, classification and recognition object are accomplished according to the computed results, furthermore, objects do not need to be trained in the paper. The algorithm is simple and the recognition rate is rather high. Moreover, the objects with noise are able to be recognized correctly. Experimental results demonstrate that the proposed algorithm is invariant to the translation, rotating and scaling of objects. So the efficiency is proved in the paper.